Before running this experiment, we identified Clarendon Dental Spa as the Leeds practice whose website appeared to match a detailed customer requirement most completely.
We then submitted the same query to Google AI Mode five times.
Clarendon was recommended in every response. It ranked first three times and second twice.
This does not prove that Clarendon’s website content caused its consistent inclusion. However, it supports our working theory that businesses may have a better chance of appearing in detailed AI recommendations when they publish clear evidence addressing the customer’s complete requirement.
Taking our earlier testing a step further
In an earlier experiment, we examined whether AI matches specific customer needs with specialist businesses.
This test takes that idea a step further.
Instead of running a query and then examining the businesses Google recommended, we first tried to identify the website that appeared to offer the strongest evidence match. We then tested whether Google AI Mode would repeatedly reach the same conclusion.
We also wanted to move away from the accountancy examples used in some of our previous tests.
A nervous dental patient provided a useful scenario because their choice of practice could depend on several connected requirements.
Someone who has avoided treatment following a bad experience is not simply looking for any nearby dentist. They may need:
- Experience helping nervous or phobic patients.
- Understanding of people returning after a long absence.
- A non-judgemental approach.
- Treatment explained clearly and undertaken at their pace.
- Conscious sedation when clinically appropriate.
- Clear prices.
- A private practice accepting new adult patients.
This gave us a realistic situation in which one business might provide a much more complete answer than another.
The query we tested
We created the following prompt:
I live in Leeds and have severe dental anxiety following a bad previous experience. I have avoided the dentist for several years and need a private dentist accepting new adult patients. I would like a practice experienced in helping nervous or phobic patients, where treatment will be explained without judgement and conscious sedation is available if clinically appropriate. Clear published fees would also be helpful. Which three Leeds dental practices should I consider, and why?
This is considerably more detailed than a traditional search such as “private dentist Leeds”.
However, it represents the kind of request someone might now make using an AI-powered search tool.
It also follows the approach described in our guide to building an AI visibility query set around real customer situations. The query was constructed around a plausible customer need rather than a keyword whose commercial relevance was unclear.
How we identified Clarendon
Before submitting the prompt to Google AI Mode, we used ChatGPT to help locate possible Leeds dental practices. We then examined the evidence published on their official websites.
The candidate-discovery process was therefore AI-assisted. We did not find Clarendon entirely independently of AI.
However, we did not use any results from the subsequent Google AI Mode experiment to choose our predicted practice. We formed our prediction first by comparing the published evidence.
Among the practices examined, Clarendon Dental Spa appeared to provide the most complete overall match.
Its sedation page is specifically titled “Sedation Dentistry for Nervous Patients” and refers directly to:
- Dental anxiety and phobic patients.
- Anxiety arising from previous dental experiences.
- Helping patients feel calm and supported.
- Allowing the patient to remain in control.
- Judgement-free care.
- Sedation delivered by trained clinicians.
- Published sedation prices starting from £200 per session.
Clarendon also describes three forms of conscious sedation:
- Intravenous sedation.
- Inhalation sedation.
- Oral sedation.
The oral option involves prescribed medication being taken before treatment in accordance with the practice’s clinical instructions. This can allow an anxious patient to begin feeling more relaxed before the appointment starts.
Other Clarendon pages published new-patient fees and addressed people who had not attended a dentist for several years.
We therefore made a prospective prediction:
If Google AI Mode successfully connects a detailed customer requirement with a business publishing closely matching evidence, Clarendon Dental Spa should appear regularly.
The five-run results
We submitted exactly the same prompt to Google AI Mode five times.
Each response recommended three Leeds dental practices.
| Practice | Run 1 | Run 2 | Run 3 | Run 4 | Run 5 | Appearances |
|---|---|---|---|---|---|---|
| Clarendon Dental Spa | 1st | 2nd | 1st | 2nd | 1st | 5/5 |
| Aesthetique Dental Care | 2nd | 1st | 2nd | — | 2nd | 4/5 |
| Infinity Dental Clinic | — | 3rd | 3rd | 3rd | 3rd | 4/5 |
| Yorkshire Dental Suite | 3rd | — | — | — | — | 1/5 |
| HQ Dental | — | — | — | 1st | — | 1/5 |
Clarendon was the only practice recommended in every response and never appeared below second place.
Aesthetique Dental Care and Infinity Dental Clinic were each recommended four times. Together, these three practices occupied 13 of the 15 available recommendation positions.
Repeating the query was important. One response would have shown only which three practices appeared on that occasion. Five runs revealed which recommendations were relatively stable and which were occasional selections.
Our guide to running a repeatable manual AI visibility test explains how businesses can conduct this type of repeated check for themselves.
Did the recurring practices genuinely match the query?
We checked the principal claims against the practices’ own websites rather than assuming that the AI-generated explanations were accurate.
| Evidence examined | Clarendon | Aesthetique | Infinity |
|---|---|---|---|
| Dedicated nervous or phobic patient content | Yes | Yes | Yes |
| Previous bad experience or long absence addressed | Yes | Yes | Yes |
| Explanation, reassurance or patient control | Yes | Yes | Yes |
| Conscious sedation | IV, inhalation and oral | IV and oral | IV |
| Named sedation clinician | Yes | Yes | Yes |
| Published sedation price | From £200 | £395 | Individual cost breakdown promised, but no price confirmed on the page checked |
| Relevant patient experiences | Supporting review evidence | Anxiety-specific reviews | Detailed phobia and trauma testimonials |
The three most frequently recommended practices all published substantial first-party evidence matching the customer’s circumstances.
Clarendon offered the strongest overall combination among the candidates we examined before the test, particularly because it described three different forms of sedation alongside dedicated support for nervous and phobic patients.
Aesthetique’s nervous-patient page contained especially detailed evidence about the experience it offers. It explained that people returning after several years would not be judged, that treatment could proceed at the patient’s pace and that patients could visit the practice before committing to treatment. It also described practical support including stop signals, music and stress balls.
Infinity’s dental-sedation page addressed previous traumatic experiences, IV conscious sedation, a full sedation assessment and a named provider with experience helping hundreds of patients with dental phobia. It also published testimonials from people describing severe dental fear and previous bad experiences.
The repeated recommendations therefore had a logical connection with the evidence available on the practices’ own websites.
Comparing the occasional and non-appearing practices
We also examined the two one-off recommendations and two relevant practices that did not appear.
| Practice | Appearances | Relevant published evidence | Apparent limitation relative to the query |
|---|---|---|---|
| Yorkshire Dental Suite | 1/5 | Phobia content, sedation options and published prices | Much of its detailed sedation evidence related to implants or complex treatment |
| HQ Dental | 1/5 | Nervous-patient reassurance and IV sedation on referral | Several detailed claims in the AI response were not found on the HQ pages checked |
| Horsforth Smile Clinic | 0/5 | Nervous-patient page, IV sedation and clear prices | Less detailed evidence about long absences, judgement and relevant patient outcomes |
| Hewitt Dental Practice | 0/5 | Accepting new patients, published prices and Dental Phobia Certified | We did not find evidence of conscious sedation |
This comparison needs to be interpreted carefully.
Non-appearance does not mean that a practice provides a poor service or would be unsuitable for an individual patient. It only means that Google AI Mode did not select it during these five runs.
Nevertheless, among the practices we examined, the three appearing most consistently also provided the most complete combination of evidence matching the different parts of the query.
Recommendation consistency did not guarantee factual accuracy
Although the overall recommendation pattern appeared logical, some details in the AI-generated explanations were wrong or poorly supported.
For example:
- Clarendon’s current new-patient examination price was £79, not the £69 quoted in several responses.
- Aesthetique’s new-patient examination was £50, not £35 or £110. Its £35 price applied to routine examinations for registered patients.
- Infinity stated that its named provider had treated hundreds of patients with dental phobia. One AI Mode response changed this to thousands.
- Several detailed claims made about HQ Dental were not found on the HQ pages we checked, while citations attached to some of those claims led to other dental practices.
AI Mode appeared to combine information from different sources when constructing some of its explanations. In the case of HQ, this created a more complete nervous-patient profile than its own pages appeared to support.
This reveals an important distinction:
A business can be a plausible recommendation while some of the reasons given for recommending it are inaccurate.
Businesses should therefore measure more than whether they appear. They should also check whether AI describes the recommended business accurately.
Does the ChatGPT-assisted discovery weaken the result?
Using ChatGPT to find potential candidates is a limitation, but it does not make the test circular in the strict sense.
A circular test would have involved running the Google AI Mode query first, noticing that Clarendon appeared repeatedly and then working backwards to describe it as our predicted practice.
That is not what happened.
ChatGPT helped us locate possible practices, but we formed our prediction from the evidence published on their websites before running Google AI Mode. It was therefore a prospective prediction rather than a retrospective explanation of the results.
However, ChatGPT and Google may be influenced by overlapping web signals. Clarendon’s reviews, general search visibility, links, brand prominence and mentions elsewhere could have helped both systems discover it.
A stronger future test could start with a predetermined list from a source such as the Care Quality Commission register. Every eligible practice could then be assessed against the same evidence checklist before consulting either ChatGPT or Google AI Mode.
For this experiment, we can say that the result supports our theory. We cannot say that Clarendon’s website content caused its five appearances.
What can businesses learn from the result?
The practical lesson extends well beyond dentistry.
A business should not simply state that it offers a particular service. It should publish evidence explaining how it meets the wider customer need surrounding that service.
For a dental practice, “nervous patients welcome” provides much less usable information than explaining:
- How the practice helps people returning after a long absence.
- How treatment is explained.
- How the patient remains in control.
- Which sedation options are available.
- Who provides the sedation.
- What the service costs.
- What patients with similar experiences say.
The same principle could apply to an accountancy firm wanting to demonstrate Xero expertise. Rather than merely claiming to be a Xero expert, it could publish its certification, identify experienced staff, describe the types of clients it supports, show examples of problems resolved and include relevant customer experiences.
The starting point is choosing AI search prompts that genuinely matter to the business. A business does not need to identify every possible long-tail wording. It can begin with a small group of realistic queries representing the customers and services it most wants to attract.
It can then ask:
- Does the business appear consistently?
- How is it described?
- Which competitors appear?
- What evidence do the recurring recommendations publish?
- Does the business’s own website clearly prove that it can meet the same need?
What this experiment shows
This was a small experiment involving one query and five Google AI Mode responses.
It does not prove that Clarendon’s website content caused its repeated inclusion. ChatGPT assisted with candidate discovery, and other web signals may also have influenced the results.
Nevertheless, we identified Clarendon prospectively as the strongest overall evidence match among the practices examined. It subsequently became the only practice recommended in all five runs and never appeared below second place.
Aesthetique and Infinity also appeared consistently, and verification showed that both published unusually detailed evidence addressing the customer’s circumstances. That makes the overall pattern difficult to dismiss as purely random, even though five runs cannot establish causation.
The practical implication is not that one comprehensive page guarantees AI visibility. It is that businesses give AI systems stronger evidence to work with when they clearly demonstrate how their services meet a specific customer need.
Our working theory therefore survives this test:
Businesses may improve their chances of appearing in detailed AI recommendations by publishing clear, specific and connected evidence addressing the customer’s complete requirement—not merely by listing the service they sell.